Extract all data for the June months over the years from a timetable

I want to pull all the data only for the month of june out of this time table but I do not want to group it all into a single bin "june". Ultimately I want to know how many days there was precipitation in June of 1948, June of 1949, ect. I have been able to use the groupsummary function to almost get there but it lumps all the years data into a single June bin. My data spans 60 years and I want 60 junes worth of data basically. I am new to using the timetables in matlab and I have read many of the help pages but cannot seem to figure out how to just pull out a single month. Thanks for any help!

whos -file preciptimedata.mat

  Name                Size             Bytes  Class        Attributes

  preciptimedata       -              352359  timetable              

Answer 1

[dpb]

Piece 'o cake -- granted, it takes a little while to get used to using grouping variables and all, but...

load https://www.mathworks.com/matlabcentral/answers/uploaded_files/1115475/preciptimedata.mat

Error using load
Unable to read file 'https://www.mathworks.com/matlabcentral/answers/uploaded_files/1115475/preciptimedata.mat'. If it is a Version 7 or earlier MAT-file, consider saving your data afresh in Version
7.3 MAT-files to access it from a remote location.

tP=preciptimedata; clear preciptimedata     % I like shorter variable names...
tP.Properties.VariableNames={'Rainfall'};   % make meaningful variable name for data
tP=addvars(tP,month(tP.precipdates),year(tP.precipdates), ...   % create grouping variables
    'NewVariableNames',{'Month','Year'},'Before','Rainfall');
tG=groupsummary(tP,{'Month','Year'},"nnz","Rainfall");      % and do the work..
tJun=tG(tG.Month==6,:)

Oh, bummer...same code from home system (R2020b) yields..

>> tJun=tG(tG.Month==6,:);
>> head(tJun)
ans =
  8×4 table
    Month    Year    GroupCount    nnz_Rainfall
    _____    ____    __________    ____________
      6      1949        30             21     
      6      1950        30             13     
      6      1951        30             24     
      6      1952        30              8     
      6      1953        30             17     
      6      1954        30             10     
      6      1955        30             16     
      6      1956        30             11     
>> 

Of course, you don't have to make new table to see June, just did here so could use head as shorthand to display. tG holds all months summary counts so can now just look up any month desired. @Walter Roberson's solution contains same data; just slightly different way to get there by using external variables and findgroups/splitapply workflow instead of sticking with the original timetable. No real advantage one way or t'other that I see although with grouping variables in the table, you've already got 'em at hand to apply with any other analysis that comes up...

[dpb]

No idea where the geographical area for these data is; clearly somewhere pretty wet (NOT SW KS). It caught my attention and is interesting, however, that the years above of '52-'56 are on average lower than the initial three average and in particular 1952 was very dry. Those years coincide with the great drought of the '50s in the US central plains; there were a couple of those years which were even drier on the family farm here than any of those during the "Dirty Thirties".

May be only coiincidence, but is perhaps reflective of global weather pattern relationships -- it is known that our dry periods are correlated/in part caused by the La Nina/El Nino cycles off the S American coast...I've often thought if had data it would be interesting to see if, indeed, the 30s were also in synch with that phenomenon. These also don't go back that far, though -- I've never found such data readily accessible although I've not done truly extensive searching.

Answer 2

[Walter Roberson]

tested...

pcts = load('preciptimedata.mat');
pct = pcts.preciptimedata;

[y,m,d] = ymd(pct.Properties.RowTimes);

junepct = pct(m == 6, :);
juneyear = y(m == 6);
G = findgroups(juneyear);

rain_days = splitapply(@(year,var1) [year(1),nnz(var1)], juneyear, junepct.Var1, G);
rain_days = timetable(rain_days(:,2), 'VariableNames', {'days of rain'}, 'RowTimes', datetime(rain_days(:,1),1,1,'Format','uuuu'));

Answer 3

[Isaac Lammers]

So I went out of town for labor day weekend and was away from service but brought my computer. Thank you all for the answers! Here is what I came up with after much trial and error. I think I will probably implement one of your solutions that are above though as they seem more elegant than a for loop haha.

In regards to the data that I have, the geographical location is Boulder, CO. I am fairly certain it is a data set from one of my professors research however I am just taking a data analysis course so I am not focused so much on what the data is, as much as I am learning to manipulate it! Thanks for the help!

My code below probably would need some of the additional variable I have defined further up in the homework set but it gets point across I think...

year = transpose(1949:2008);
for i = 1:1:size(year) %range of years of data
    date1 = datetime(year(i),06,01); %Method for pulling only the june data
    date2 = datetime(year(i),07,01);
    TR = timerange(date1,date2);
    precipmonthjune = preciptimedata(TR,:);
    yearjuneprecip = groupsummary(precipmonthjune, 'precipdates','year','nnz');
    yearjuneprecip = table2array(yearjuneprecip(:,3));
    juneprecipdays(i,:) = yearjuneprecip;
end

[Walter Roberson]

numel(year) not size(year) because size with a single input always returns a vector.

[dpb]

"the geographical location is Boulder, CO"

AHA! Not all that far away from us...but close enough to the mountains to have the summer monsoons that we do not have the luxury of...but also probably affected somewhat similarly and would find fairly high correlation with our area over time.